Conversation
Council of AI · measure, sign, verify
Council of AI governed workspace
What should the Council help you do?
Ask once, then keep the conversation, selected tools and evidence together. The composer below is the control point; starter actions only prefill or open a real surface.
Inspect a model arenaRecorded rounds and deterministic grading; never a pretend live battle.Train on a GSPC axisLearn, play, explain, propose a fix, then stop for human review.Verify evidenceRecompute a card hash and signature in the browser.Diagnose and draft a fixPrefills the Council—not sent, executed or approved automatically.
MCP tool catalogue
13 declared by tools/listOpen allCatalogue discovery is live. A tool becomes runtime-observed only after its own tools/call completes.
Account overview and recent measurements
Risk Assessment
Evaluate your AI systems against global safety frameworks
Want a signed result in 2 minutes instead?
Run the free EU AI Act risk check and get a cryptographically signed, independently verifiable verdict →
EU AI Act Assessment
Comprehensive risk classification per Regulation 2024/1689
113 articles•~45 min
NIST AI RMF Assessment
Four-function framework: GOVERN, MAP, MEASURE, MANAGE
7 characteristics•~30 min
TC260 Assessment
China AI Safety Governance Framework 2.0 compliance
14 measures•~35 min
Multi-Framework Assessment
Combined assessment across all supported frameworks
All frameworks•~90 min
Answers from published measurement, or it refuses.
